The performance demands in modern engineering applications have motivated the use of complex signal processing and control algorithms. This in turn has put constraints on computing capabilities of processors. Thus, to achieve efficient implementation of an algorithm a matching of the computing capabilities of processors with the computing requirements of the algorithm is required. This paper presents an investigation into the characteristic behaviour of algorithms for efficient real-time implementation using heterogeneous parallel computing. Accordingly, several characteristic models describing categories of algorithms are considered and a generalised mathematical model for task to processor allocation in a heterogeneous computing framework is developed and demonstrated.